Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Video vehicle detection method and counting method based on adversarial network learning

A technology of network learning and vehicle detection, applied in the field of image detection, can solve the problems of intelligent transportation, complex image background, complex background, etc., and achieve the effect of solving the difficulty of feature acquisition

Inactive Publication Date: 2018-01-09
ANHUI SUN CREATE ELECTRONICS
View PDF2 Cites 19 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0002] The detection and automatic counting of vehicles in video has always troubled the traffic analysis and statistics of intelligent transportation. Due to the different types and shapes of vehicles in the video environment, and the images in the video are seriously disturbed by various noises, large-scale vehicle statistical analysis cannot be completed manually
The current video vehicle detection and counting work mainly relies on conventional machine learning algorithms, but the image background in the video is complex and the scale of processing information is large, making it difficult to accurately analyze and count video vehicles
[0003] At the same time, due to the complex background of the vehicle video image in the video, the influence of appearance and posture, the traditional automatic vehicle detection and counting method is inefficient and poor in robustness, and it can only exist in the experimental stage and is difficult to apply in practice.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Video vehicle detection method and counting method based on adversarial network learning
  • Video vehicle detection method and counting method based on adversarial network learning
  • Video vehicle detection method and counting method based on adversarial network learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0050] Such as Figure 1-2 As shown, a vehicle detection method based on adversarial network learning video, including the following steps:

[0051] S1. Acquisition of vehicle video images. In the existing project "Smart City", several vehicle video images are collected as training images, and the focus of the collected vehicle video images is on the vehicle, including the front, rear and side of the vehicle. The training images are normalized to obtain multiple normalized vehicle video images, which are used as vehicle video image training samples. In order to facilitate the rapid detection and calculation of the computer and reduce the computational complexity, the size of all training images is normalized to 32×32 pixels.

[0052] S2. Obtain a vehicle video image detection model; perform construction and confrontation training on the video image discrimination network D(x,l) and video image generation network G(z,l) under conditional constraints; in addition, obtain vehic...

Embodiment 2

[0078] A method for counting vehicles based on a video of confrontational network learning, implementing the method steps of vehicle detection in a video based on confrontational network learning in Embodiment 1, since the detection is for each 32×32 image block as input, for a The image has been divided into image blocks, so it is possible to realize counting statistics for multiple vehicles in one image. After implementing the steps in Example 1, continue to implement the following steps:

[0079]Assume that for the detected vehicle video image, there are N, and N is the number of vehicles predicted in the detected video image, which is equivalent to the detection result that the pixel size of the vehicle video image training sample is the size of the image block. In this embodiment The normalized pixel size is 32×32 pixels, and the area of ​​each detection result in the original image is marked as R 1 , R 2 ...,R N , the formula for calculating the number of vehicles is a...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a video vehicle detection method and a counting method based on adversarial network learning. The vehicle detection method comprises the following steps: S1) obtaining a vehicle video image; S2) obtaining a vehicle video image detection model; S3) obtaining a video image of a vehicle to be detected; and S4) marking specific position of the vehicle in the vehicle video image. The advantages are that through setting of a video image generation network G (z,l), a lot of vehicle video image training samples are increased, and the problem of difficult feature acquisition under complex vehicle conditions is solved; and through adversarial training, identification ability of a video image discrimination network D(x, l) is improved, and accurate location detection of the vehicle is realized.

Description

technical field [0001] The invention relates to the technical field of image detection, in particular to a vehicle detection method and a counting method based on adversarial network learning video. Background technique [0002] The detection and automatic counting of vehicles in video has always troubled the traffic analysis and statistics of intelligent transportation. Due to the different types and shapes of vehicles in the video environment, and the images in the video are seriously disturbed by various noises, large-scale vehicle statistical analysis cannot be completed manually . The current detection and counting of video vehicles is mainly done by conventional machine learning algorithms, but the image background in the video is complex and the scale of processed information is large, making it difficult to accurately analyze and count video vehicles. [0003] At the same time, due to the complex background of the vehicle video image in the video, the influence of a...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/00G06K9/62
Inventor 任子晖李铮张丽君张兆义高洪昌胡俊孙林
Owner ANHUI SUN CREATE ELECTRONICS
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com